6213

2009, 113, 6213–6216 Published on Web 05/13/2009

Limits on Fluorescence Detected Circular of Single Helicene Molecules

Yiqiao Tang,† Timothy A. Cook,‡,§ and Adam E. Cohen*,†,‡ Department of Physics, HarVard UniVersity, Cambridge, Massachusetts 02138, and Department of Chemistry and Chemical Biology, HarVard UniVersity, Cambridge, Massachusetts 02138 ReceiVed: April 19, 2009; ReVised Manuscript ReceiVed: May 6, 2009

Fluorescent imaging of single helicene molecules is applied to study the optical activity of chiral fluorophores. In contrast to the previous report by Hassey et al. (Science 2006, 314, 1437), the dissymmetry factors of single chiral fluorophores are found not to differ significantly from the bulk value of |g| < 10-4 at 457 nm. Linear dichroism and birefringence of the dichroic inside the fluorescence microscope change the state of the incoming laser beam significantly; i.e., circular polarized sent into the microscope becomes highly elliptically polarized after reflection from the dichroic mirror. Compensation for this effect should be made to avoid artifacts brought by linear dichroism in single immobilized molecules.

Single-molecule measurements have been used to probe the H )-µ · E - θ:∇E - m · B states and dynamics of a huge variety of biophysical and condensed matter systems,2-4 and often yield information that where µ is the electric dipole operator, θ is the electric is missed by bulk, ensemble-averaged techniques. In particular, quadrupole operator, and m is the magnetic dipole operator; E measurements at ambient and low temperatures provide detailed is the electric field, and B is the magnetic field. The rate of information about the interactions between single fluorescent excitation from an initial state i to a final state f involves the 2 molecules and their local environment. The heterogeneously expression |〈f|H|i〉| , as specified by Fermi’s golden rule. This 2 broadened lines visible in bulk often resolve into much narrower expression contains a term proportional to |µif| , as well as cross- single-molecule lines, which show spectral diffusion,5,6 blinking,7 terms containing µ and θ or µ and m (the remaining terms are 2 and polarization fluctuations8,9 that indicate complex interactions small enough to be neglected). The term containing |µif| with the host. represents electric dipole absorption, and is not sensitive to molecular . The signs of the two cross-terms, however, Chiral molecules of a single enantiomer show differential depend on molecular chirality, so these terms are responsible absorption of left and right circularly polarized light (CPL). The for . Both cross-terms depend on the orienta- degree of circular dichroism (CD) at a particular frequency is tion of the molecule relative to the incident field. Upon averaging measured by the g-value: over all molecular orientations, the electric quadrupole contribu- tion to the differential absorption averages to zero while the Publication Date (Web): May 13, 2009 | doi: 10.1021/jp903598t 13 + - - magnetic dipole term remains. Thus, electric quadrupole ≡ 2(Γ Γ ) transitions do not contribute to bulk CD, while magnetic dipole Downloaded by HARVARD UNIV on September 3, 2009 | http://pubs.acs.org g + - (Γ + Γ ) transitions do contribute. In light of the increased information available from single- molecule measurements, it is interesting to compare the chi- ( where Γ is the rate of excitation in right (+) or left (-) CPL. roptical response of a single chiral molecule to the ensemble- The g-value varies between -2 e g e 2. For most small organic averaged response. One can measure chiroptical effects in small - molecules, peptides, nucleic acids, and sugars, |g| < 10 3 in the samples by using fluorescent chiral molecules.15,16 The rate of visible.10 The smallness of g arises from the small size of fluorescence emission is proportional to the rate of excitation, molecules relative to the helical pitch of CPL: the electromag- and thus, fluorescence detected circular dichroism (FDCD) netic field undergoes a nearly imperceptible twist over a distance provides a possible probe of chiroptical effects at the single- of molecular dimensions. molecule level. Single-molecule FDCD might differ from bulk At the quantum mechanical level, CD arises through an FDCD for several reasons: interference between electric dipole and either electric quadru- (1) Electric quadrupole and magnetic dipole matrix elements pole or magnetic dipole transitions.11 The Hamiltonian relevant of orientationally fixed single molecules may differ from their 13,17 to CD for a chiral molecule in an electromagnetic field is:12-14 rotationally averaged values in bulk systems. (2) Local interactions with the host could modify the oscillator * To whom correspondence should be addressed. E-mail: cohen@ strengths or frequencies of the electric quadrupole or magnetic chemistry.harvard.edu. dipole transitions, to induce molecule-to-molecule variations in † Department of Physics. CD, i.e., heterogeneous broadening of CD lines. ‡ Department of Chemistry and Chemical Biology. § Current address: School of Law, University of Virginia, Charlottesville, The smallness of the electric quadrupole and magnetic dipole VA 22903. contributions to the Hamiltonian is purely geometrical in origin, 10.1021/jp903598t CCC: $40.75  2009 American Chemical Society 6214 J. Phys. Chem. A, Vol. 113, No. 22, 2009 Letters

a consequence of the small size of a molecule relative to the wavelength of light. Orientational averaging reduces the strength of the electric dipole/magnetic dipole interference by a factor of 3 relative to its peak value.13 Electric quadrupole transitions average to zero in bulk, so bulk measurements provide no guidance on their magnitude. Geometrical considerations, however, indicate that the electric quadrupole term should be of the same order of magnitude as the magnetic dipole term. Density functional simulations of single helicene molecules support the prediction that single-molecule g-values are of the order of 10-3, of the same order as in bulk at the same excitation wavelength.18 Nonetheless, theoretical models may be wrong, so there was considerable interest in the report from Hassey et al.1 that single molecules of bridged triarylamine helicenes show g-values ranging from +2to- 2, while in bulk the two enantiomers have |g| < 10-4 at a probe wavelength of 457 nm. Furthermore, they reported that both enantiomers of this compound show very similar broad distributions of single-molecule g-values, and that the wavelength dependence of the single-molecule g-values does not correspond to the bulk, even when averaged over many single molecules.18 We attempted to replicate the experiments of Hassey et al. and initially saw similar broad distributions of g-values. However, after correcting for the linear and linear dichroism inherent to the dichroic used in single- molecule FDCD experiments, the distribution of g-values collapses to a sharp distribution around g ) 0. The uncertainty in the measurements does not allow us to distinguish the g-values of opposite enantiomers. Figure 1. Bridged triarylamine helicenes. (a) Structure of the P-2 enantiomer. The camphanate group is used as a chiral resolving agent Methods and renders the M-2 and P-2 species technically diastereomers. However, the camphanate group has no detectable effect on the optical Triarylamine helicenes M-2 and P-2 (Figure 1a) were properties of either species. (b) Circular dichroism spectrum of M-2 synthesized following literature procedures19,20 with modifica- and P-2, measured with a JASCO CD spectrometer. The spectrum of tions. (2-Methoxyphenyl)-naphthalenylamine was synthesized the P-2 was scaled by a constant factor to correct for a slight difference in concentration between the two samples. using (1,1′-bis(diphenylphosphino)ferrocene)PdCl2 · CH2Cl2 as the amination catalyst,21 and deprotection of the helicene methyl 22 ether was accomplished using BBr3. The molecules we (FM600QM Thorlabs). The LCVR is driven with a 2 kHz square prepared are identical to the ones Hassey et al. used in ref 1. wave. An arbitrary state of ellipticity is generated by adjusting The products were characterized by 1H NMR, 13C NMR, and the LCVR drive voltage. We have mirrors between the Publication Date (Web): May 13, 2009 | doi: 10.1021/jp903598t mass spectrometry; all data match the literature spectra. The and the microscope, to avoid introducing spurious phase shifts circular dichroism (Figure 1b) and fluorescence (not shown) onto the beam. The only reflection after the initial polarizer is

Downloaded by HARVARD UNIV on September 3, 2009 | http://pubs.acs.org spectra are identical to earlier reports.19 Both enantiomers show off the dichroic mirror. strong FDCD in bulk when excited with light at 355 nm, with A lens with a focal length of 15 cm (LA1433-A Thorlabs) g-values of +8.0 × 10-3 for M-2 and -8.1 × 10-3 for P-2. At brings the light to a focus at the back focal plane of the objective. 457 nm, the wavelength used in the experiments of Hassey et After passing through the lens, the light is deflected upward by al., the g-values are |g| < 10-4. a dichroic mirror. We performed experiments using either of Following the procedure of Hassey et al., we made solutions two dichroics: DC1, 460 nm long pass; DC2 (Olympus DM500), of the helicenes at 10-8 M in methanol and drop-cast these 500 nm long pass. Our objective lens is a 60×, N.A. 1.45, oil solutions onto a Zeonor film (ZF-14 ZEONEX/ZEONOR). A immersion, plan apochromat (1-U2B616 Olympus). Fluores- fused silica coverslip was cleaned in Piranha solution (3:1 cence from the sample is collected by the same objective, passed H2SO4:H2O2, Caution: highly corrosive). The Zeonor film was through the dichroic mirror, and separated from the excitation placed, helicene side down, onto the coverslip, and the samples light by a 470 nm long-pass emission filter (HQ470LP Chroma). were imaged in an inverted single-molecule fluorescence Images are collected on an Andor iXon+ electron-multiplying microscope. CCD (DU-897E-CS0-#BV), cooled to -50 °C. The optical setup is shown in Figure 2a. The setup is built A program written in LabView synchronizes the acquisition around an Olympus IX71 inverted fluorescence microscope. The of images with the application of voltages to the LCVR. The light source is a Melles Griot argon-ion laser operating at 457 LCVR has an ∼30 ms response time to a change in the nm (0.4 mW at the sample). A narrow-band excitation filter amplitude of its driving voltage, so each image acquisition starts (D457/10x Chroma) is used to remove plasma emission at other 100 ms after a change in the voltage on the LCVR. In a typical wavelengths. The light is polarized by a Glan Thompson experiment, 40 images are acquired, at 400 ms exposure per polarizer (10GL08 Newport), passed through a frame, with a switch in the polarization state between each variable retarder (LCVR, LRC - 200 - VIS, Meadowlark frame. An amplified photodiode detector (PDA36A, Thorlabs) ), and then through a quarter wave Fresnel rhomb retarder is mounted on top of the microscope to monitor the intensity Letters J. Phys. Chem. A, Vol. 113, No. 22, 2009 6215 of the transmitted laser beam. This photodetector allows us to 1 (1 ) correct for laser power fluctuations. 1.25 (0.75e0.78i Prior to each experiment, we measure the polarization state of the light entering the microscope and emerging from the objective. A polarizer (GL10 Thorlabs) is temporarily placed for DC1 and in the beam path at the point of measurement. The transmitted intensity is recorded using a power meter (FieldMaxII-TO 1 1 Coherent Inc.) as a function of the angle of the polarizer. (( 1.18i ) Data are acquired under two conditions for each enantiomer: 1.43 1.03e alternating left- and right-CPL sent into the microscope and alternating left- and right-CPL at the sample plane. The analysis for DC2. These polarizations are generated by applying ap- for all movies is completely automated using custom software propriate voltages to the LCVR. Measurements confirmed written in Matlab, with identical parameters for all data sets. ellipticities >96% above the sample for both right- and left- Thus, there is no possibility for bias in selecting molecules or CPL. extracting their intensities. Figure 3b shows a histogram of g-values measured for single molecules with CPL before the microscope (using DC1) and Results with CPL at the sample (using DC2). The apparent g-values Figure 2b shows an image of single helicene molecules on are broadly distributed in the case where the CPL is generated the Zeonor film. We first imaged the molecules with alternating before the microscope, using either DC2 (data not shown) or left- and right-CPL sent into the microscope. For both polariza- DC1. In contrast, with CPL at the sample, the g-values are not ) ( tions, the ellipticity before the microscope is greater than 96%. significantly different from zero: g 0.026 0.27 for M-2 ) ( Many molecules show strong asymmetries in their brightness, and g 0.033 0.29 for P-2. The width of the observed as shown in Figure 3b, leading to apparent g-values ranging distribution is largely due to statistical uncertainties in the from -2 to 2. However, under these conditions the light extraction of single-molecule g-values in the presence of shot- emerging from the microscope is not circularly polarized. noise and molecular blinking and photobleaching. Thus, current Ellipticities are 36% for DC1 and 55% for DC2. experimental techniques cannot detect deviations from 0 in the To generate CPL at the sample, we measured the Jones g-values of single helicene molecules measured at 457 nm. matrices23 for the dichroics. They are Discussion 10 We have shown that the broad distribution of g-values observed by Hassey et al. can be explained by linear dichroism24 (0 0.79i ) 1.34e in the randomly oriented helicene molecules, coupled with imperfect of the illumination. We believe and that such a mechanism is a more probable explanation for their data than is anomalously large circular dichroism at the single- (10 ) molecule level, which would require a hitherto unknown 0 0.97e0.39i physical effect. In addition to the dichroic mirrors discussed above, a third dichroic mirror with a cutoff of 570 nm (Olympus for DC1 and DC2, respectively. We solved the matrix equations DM570) also shows strong linear birefringence for 532 nm light. to identify the Jones vectors of the input polarization states We conclude that strong linear birefringence is a general feature

Publication Date (Web): May 13, 2009 | doi: 10.1021/jp903598t which lead to CPL at the sample. These are of dichroic mirrors. Finding a dichroic with nearly equal Downloaded by HARVARD UNIV on September 3, 2009 | http://pubs.acs.org

Figure 2. Imaging of single helicene molecules under light of controlled polarization. (a) Schematic diagram of optical setup for wide-field imaging. See text for a detailed description. (b) Image of single helicene molecules. A diffuse background image, calculated using a 2-dimensional median filter, was subtracted from the raw image to enhance the contrast of the single molecules. 6216 J. Phys. Chem. A, Vol. 113, No. 22, 2009 Letters

reflective elements in their optical train and so were not subject to this source of error.25 Circular dichroism measurements at the single-molecule level promise important information on molecular structure and local environmental interactions. However, uncompensated linear birefringence and linear dichroism in optical elements inside a microscope can cause linear dichroism to masquerade as circular dichroism. Caution is advised in performing and interpreting such experiments. Acknowledgment. We thank Prashant Jain and Sijia Lu for helpful discussions, Charles Lieber and Sunney Xie for loans of equipment, and Gregory Verdine for use of the CD spectrometer. We thank Michael Barnes for comments on an early draft of this manuscript. This work was partially supported by a Dreyfus New Faculty Award and the MITRE Corporation and the U.S. Government’s Nano-Enabled Technology Initiative. References and Notes (1) Hassey, R.; Swain, E. J.; Hammer, N. I.; Venkataraman, D.; Barnes, M. D. Science 2006, 314, 1437. (2) Moerner, W. E.; Kador, L. Phys. ReV. Lett. 1989, 62, 2535. (3) Moerner, W. E. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 12596. (4) Selvin, P. R.; Ha, T. Single Molecule Techniques: A Laboratory Manual; Cold Spring Harbor Laboratory Press: New York, 2007. (5) Bauer, M.; Kador, L. J. Chem. Phys. 2004, 120, 10278. (6) Vainer, Y. G.; Naumov, A. V. Opt. Spectrosc. 2005, 98, 747. (7) Dickson, R. M.; Cubitt, A. B.; Tsien, R. Y.; Moerner, W. E. Nature 1997, 388, 355. (8) Le Floc’h, V.; Brasselet, S.; Roch, J. F.; Zyss, J. J. Phys. Chem. B 2003, 107, 12403. (9) Dickson, R. M.; Norris, D. J.; Moerner, W. E. Phys. ReV. Lett. Figure 3. Fluorescence of single helicene molecules in CPL. (a) Time- 1998, 81, 5322. traces of single helicene molecules with CPL sent into the microscope (10) Inoue, Y.; Ramamurthy, V. Chiral Photochemistry (Molecular and Supramolecular Photochemistry); CRC Press: New York, 2004. (top) and CPL generated at the sample plane (bottom). There is a large V apparent g-value when CPL is sent into the microscope but not when (11) Schellman, J. A. Chem. Re . 1975, 75, 323. (12) Barron, L. D. Molecular Light Scattering and Optical ActiVity; CPL is generated at the sample. Both molecules show single-step Cambridge University Press: Cambridge, U.K., 2004. photobleaching. (b) Histogram of g-values for both enantiomers, under (13) Craig, D. P.; Thirunamachandran, T. Molecular Quantum Electro- both polarization conditions. With CPL at the sample, the g-values are dynamics; Dover Publications: New York, 1998. much more narrowly distributed. The number of molecules measured (14) Disch, R. L.; Sverdlik, D. I. Anal. Chem. 1969, 41, 82. for each histogram are the following: M-2 CPL in, 303; P-2 CPL in, (15) Tinoco Jr, I.; Turner, D. H. J. Am. Chem. Soc. 1976, 98, 6453. 173; M-2 CPL out, 169; P-2 CPL out, 234. (16) Turner, D. H.; Tinoco Jr, I.; Maestre, M. J. Am. Chem. Soc. 1974, 96, 4340. (17) Tinoco Jr, I.; Ehrenberg, B.; Steinberg, I. Z. J. Chem. Phys. 1977, 66, 916. reflectivities for s- and p-polarized light (such as our dichroic (18) Hassey, R.; McCarthy, K. D.; Swain, E.; Basak, D.; Venkataraman, DC2) is not sufficient to preserve the polarization state of D.; Barnes, M. D. Chirality 2008, 20, 1039. Publication Date (Web): May 13, 2009 | doi: 10.1021/jp903598t incident light; the phase shift between s- and p-polarizations is (19) Field, J. E.; Muller, G.; Riehl, J. P.; Venkataraman, D. J. Am. Chem. Soc. 2003, 125, 11808. important too. An observation of CPL in light back-reflected (20) Field, J. E.; Hill, T. J.; Venkataraman, D.; He, H.; Janso, J. E.; Downloaded by HARVARD UNIV on September 3, 2009 | http://pubs.acs.org from the sample is also not proof of CPL at the sample, because Williamson, R. T.; Yang, H. Y.; Carter, G. T.; Carra, C.; Ghigo, G. J. Org. the back-reflected light has undergone a second phase-shifting Chem. 2003, 68, 6071. reflection at the dichroic. (21) Driver, M. S.; Hartwig, J. F. J. Am. Chem. Soc. 1996, 118, 7217. (22) Ten equivalents of BBr (1.0MinCHCl ) were added to a solution If CPL is sent into a fluorescence microscope, the dichroic 3 2 2 of the helicene methyl ether in CH2Cl2 at 0 °C. The reaction was stirred at mirror converts this light into elliptically polarized light at the 25 °C for 18 h, quenched with saturated aqueous NaHCO3, and worked up sample, with different principal axes arising from the two input as in ref 20. Yield was quantitative. (23) Schellman, J.; Jensen, H. P. Chem. ReV. 1987, 87, 1359. circular polarizations. The linearly polarized component of this (24) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Springer illumination leads to different rates of pure electric dipole Science: Singapore, 2006. excitation for each randomly oriented molecule. Many authors (25) Claborn, K.; Puklin-Faucher, E.; Kurimoto, M.; Kaminsky, W.; have documented the perils of performing microscopic CD Kahr, B. J. Am. Chem. Soc. 2003, 125, 14825. measurements.14,23,25 Claborn and co-workers avoided any JP903598T